Prediction of recurrence-free survival in patients with renal cell carcinoma and tumor thrombosis of the renal and inferior vena cava of levels I–II using an extended Cox model and machine learning methods

Abstract

Aim – to compare the predictive accuracy of Cox regression and machine learning (ML) methods regarding recurrence-free survival in patients with locally advanced renal cell carcinoma after radical treatment. Additionally, to investigate an extended Cox model in which the risk function is formed using a neural network approximator (DeepSurv). Material and methods. This study conducted a retrospective analysis of data from patients diagnosed with renal cell carcinoma who underwent radical nephrectomy with thrombectomy from the renal and inferior vena cava between 2007 and 2024 at the Federal State Budgetary Institution “RSC for Radiology and Surgical Technologies named after Academician A.M. Granov”. The study included 100 patients (54 men and 46 women). The median age was 61.5 years (IQR: 59.7–63). Of the total observations, disease progression was recorded in 41 cases, while in the remaining 59 cases, the data were censored. The models were evaluated based on the concordance index (C-index) and interpreted using SHAP analysis. Results. The DeepSurv neural network model demonstrated higher predictive accuracy on the test dataset compared to the classical Cox model (C-index: 0.8056 vs. 0.7917, respectively). This indicates a superior ability of DeepSurv to rank patients by individual risk of disease progression. Using SHAP analysis, the key predictors contributing most significantly to the prognosis were identified: tumor size, ISUP grade, level of tumor thrombosis, and histological tumor type. The DeepSurv model enabled the capture of complex nonlinear interactions between features, thereby improving both the interpretability and clinical applicability of the results. Conclusion. The obtained data confirm the feasibility of using machine learning methods for personalized prognosis and optimization of monitoring strategies in patients with RCC.

Ähnliche Arbeiten